Mattijs Elschot

Sachin Dudda Nagaraju, Ashkan Moradi, Bendik Skarre Abrahamsen et al.

Artificial intelligence has emerged as a transformative tool in medical image analysis, yet developing robust and generalizable segmentation models remains difficult due to fragmented, privacy-constrained imaging data siloed across institutions. While federated learning (FL) enables collaborative model training without centralizing data, cross-modality domain shifts pose a critical challenge, particularly when models trained on one modality fail to generalize to another. Many existing solutions require paired multimodal data per patient or rely on complex architectures, both of which are impractical in real clinical settings. In this work, we consider a realistic FL scenario where each client holds single-modality data (CT or MRI), and systematically investigate augmentation strategies for cross-modality generalization. Using abdominal organ segmentation and whole-heart segmentation as representative multi-class and binary segmentation benchmarks, we evaluate convolution-based spatial augmentation, frequency-domain manipulation, domain-specific normalization, and global intensity nonlinear (GIN) augmentation. Our results show that GIN consistently outperforms alternatives in both centralized and federated settings by simulating cross-modality appearance variations while preserving anatomical structure. For the pancreas, Dice score improved from 0.073 to 0.437, a 498% gain. Our federated approach achieves 93-98% of centralized training accuracy, demonstrating strong cross-modality generalization without compromising data privacy, pointing toward feasible federated AI deployment across diverse healthcare systems.

Sachin Dudda Nagaraju, Ashkan Moradi, Bendik Skarre Abrahamsen et al.

Medical image segmentation plays a crucial role in AI-assisted diagnostics, surgical planning, and treatment monitoring. Accurate and robust segmentation models are essential for enabling reliable, data-driven clinical decision making across diverse imaging modalities. Given the inherent variability in image characteristics across modalities, developing a unified model capable of generalizing effectively to multiple modalities would be highly beneficial. This model could streamline clinical workflows and reduce the need for modality-specific training. However, real-world deployment faces major challenges, including data scarcity, domain shift between modalities (e.g., CT vs. MRI), and privacy restrictions that prevent data sharing. To address these issues, we propose FedGIN, a Federated Learning (FL) framework that enables multimodal organ segmentation without sharing raw patient data. Our method integrates a lightweight Global Intensity Non-linear (GIN) augmentation module that harmonizes modality-specific intensity distributions during local training. We evaluated FedGIN using two types of datasets: an imputed dataset and a complete dataset. In the limited dataset scenario, the model was initially trained using only MRI data, and CT data was added to assess its performance improvements. In the complete dataset scenario, both MRI and CT data were fully utilized for training on all clients. In the limited-data scenario, FedGIN achieved a 12 to 18% improvement in 3D Dice scores on MRI test cases compared to FL without GIN and consistently outperformed local baselines. In the complete dataset scenario, FedGIN demonstrated near-centralized performance, with a 30% Dice score improvement over the MRI-only baseline and a 10% improvement over the CT-only baseline, highlighting its strong cross-modality generalization under privacy constraints.

Anindo Saha, Joeran S. Bosma, Jasper J. Twilt et al.

In this intercontinental, confirmatory study, we include a retrospective cohort of 22,481 MRI examinations (21,288 patients; 46 cities in 22 countries) to train and externally validate the PI-CAI-2B model, i.e., an efficient, next-generation iteration of the state-of-the-art AI system that was developed for detecting Gleason grade group $\geq$2 prostate cancer on MRI during the PI-CAI study. Of these examinations, 20,471 cases (19,278 patients; 26 cities in 14 countries) from two EU Horizon projects (ProCAncer-I, COMFORT) and 12 independent centers based in Europe, North America, Asia and Africa, are used for training and internal testing. Additionally, 2010 cases (2010 patients; 20 external cities in 12 countries) from population-based screening (STHLM3-MRI, IP1-PROSTAGRAM trials) and primary diagnostic settings (PRIME trial) based in Europe, North and South Americas, Asia and Australia, are used for external testing. Primary endpoint is the proportion of AI-based assessments in agreement with the standard of care diagnoses (i.e., clinical assessments made by expert uropathologists on histopathology, if available, or at least two expert urogenital radiologists in consensus; with access to patient history and peer consultation) in the detection of Gleason grade group $\geq$2 prostate cancer within the external testing cohorts. Our statistical analysis plan is prespecified with a hypothesis of diagnostic interchangeability to the standard of care at the PI-RADS $\geq$3 (primary diagnosis) or $\geq$4 (screening) cut-off, considering an absolute margin of 0.05 and reader estimates derived from the PI-CAI observer study (62 radiologists reading 400 cases). Secondary measures comprise the area under the receiver operating characteristic curve (AUROC) of the AI system stratified by imaging quality, patient age and patient ethnicity to identify underlying biases (if any).